Image of Transfer learning and single-polarized SAR image preprocessing for oil spill detection

Text

Transfer learning and single-polarized SAR image preprocessing for oil spill detection

This study addresses the challenge of oil spill detection using Synthetic Aperture Radar (SAR) satellite imagery, employing deep learning techniques to improve accuracy and efficiency. We investigated the effectiveness of various neural network architectures and encoders for this task, focusing on scenarios with limited training data. The research problem centered on enhancing feature extraction from single-channel SAR data to improve oil spill detection performance.
Our methodology involved developing a novel preprocessing pipeline that converts single-channel SAR data into a three-channel RGB representation. The preprocessing technique normalizes SAR intensity values and encodes extracted features into RGB channels.
Through an experiment, we have shown that a combination of the LinkNet with an EfficientNet-B4 is superior to pairs of other well-known architectures and encoders.
Quantitative evaluation revealed a significant improvement in F1-score of 0.064 compared to traditional dB-scale preprocessing methods. Qualitative assessment on independent SAR scenes from the Mediterranean Sea demonstrated better detection capabilities, albeit with increased sensitivity to look-alike.
We conclude that our proposed preprocessing technique shows promise for enhancing automatic oil spill segmentation from SAR imagery. The study contributes to advancing oil spill detection methods, with potential implications for environmental monitoring and marine ecosystem protection.

Ketersediaan
76621.3678Perpustakaan BIG (Eksternal Harddisk)Tersedia
Informasi Detail
Judul Seri
ISPRS Open Journal of Photogrammetry and Remote Sensing
No. Panggil
621.3678
Penerbit
Amsterdam : Elsevier.,
Deskripsi Fisik
8 hlm PDF, 1.441 KB
Bahasa
Inggris
ISBN/ISSN
1872-8235
Klasifikasi
621.3678
Tipe Isi
text
Tipe Media
-
Tipe Pembawa
-
Edisi
Vol.15, January 2025
Subjek
Deep learning
Transfer learning
Oil spill detection
Synthetic aperture radar (SAR)
Image preprocessing
Info Detail Spesifik
-
Pernyataan Tanggungjawab
Versi lain/terkait

Tidak tersedia versi lain

Lampiran Berkas
  • Transfer learning and single-polarized SAR image preprocessing for oil spill detection
    This study addresses the challenge of oil spill detection using Synthetic Aperture Radar (SAR) satellite imagery, employing deep learning techniques to improve accuracy and efficiency. We investigated the effectiveness of various neural network architectures and encoders for this task, focusing on scenarios with limited training data. The research problem centered on enhancing feature extraction from single-channel SAR data to improve oil spill detection performance. Our methodology involved developing a novel preprocessing pipeline that converts single-channel SAR data into a three-channel RGB representation. The preprocessing technique normalizes SAR intensity values and encodes extracted features into RGB channels. Through an experiment, we have shown that a combination of the LinkNet with an EfficientNet-B4 is superior to pairs of other well-known architectures and encoders. Quantitative evaluation revealed a significant improvement in F1-score of 0.064 compared to traditional dB-scale preprocessing methods. Qualitative assessment on independent SAR scenes from the Mediterranean Sea demonstrated better detection capabilities, albeit with increased sensitivity to look-alike. We conclude that our proposed preprocessing technique shows promise for enhancing automatic oil spill segmentation from SAR imagery. The study contributes to advancing oil spill detection methods, with potential implications for environmental monitoring and marine ecosystem protection.
Komentar
Anda harus masuk sebelum memberikan komentar